Abstract
Motivated by the desire to shift workload during periods of overload, we extend established square-root capacity sizing rules for many-server systems in the Quality-and-Efficiency Driven (QED) regime. We propose Delayed Workload Shifting (DWS) which has two defining features: when there are n users in the system, newly arriving users are no longer admitted directly. Instead, these users will reattempt getting access after a stochastic delay until they are successful. The goal of DWS is to release pressure from the system during overloaded periods, and indeed we show that the performance gain can be substantial. We derive nontrivial corrections to classical QED approximations to account for DWS, and leverage these to control stationary and time-varying system behavior.
- J. Artelejo and A. Gómez-Corral. Retrial Queueing Systems. Springer-Verlag, 2008.Google ScholarCross Ref
- F. Avram, A. Janssen, and J. van Leeuwaarden. Loss systems with slow retrials in the Halfin-Whitt regime. Adv. Appl. Prob., 45:274--294, 2013.Google ScholarCross Ref
- S. Borst, A. Mandelbaum, and M. Reiman. Dimensioning large call centers. Oper. Res., 52, 2004. Google ScholarDigital Library
- G. Falin and J. Templeton. Retrial Queues. Chapman & Hall, 1997.Google ScholarCross Ref
- S. Halfin and W. Whitt. Heavy-traffic limits for queues with many exponential servers. Oper. Res., 29(3):567--588, 1981. Google ScholarDigital Library
- O. Jennings, A. Mandelbaum, W. Massey, and W. Whitt. Server staffing to meet time-varying demand. Manag. Sci., 42(10):1383--1394, 1996.Google ScholarDigital Library
- W. Massey and R. Wallace. An asymptotically optimal design of the M/M/c/k queue. Unpublished, 2004.Google Scholar
Recommendations
Customer Abandonment in Many-Server Queues
We study G/G/n + GI queues in which customer patience times are independent, identically distributed following a general distribution. When a customer's waiting time in queue exceeds his patience time, the customer abandons the system without service. ...
State Space Collapse in Many-Server Diffusion Limits of Parallel Server Systems
We consider a class of queueing systems that consist of server pools in parallel and multiple customer classes. Customer service times are assumed to be exponentially distributed. We study the asymptotic behavior of these queueing systems in a heavy ...
Comments